Evaluation of High Temporal and Spatial Resolution 17O-MRI
Sebastian C. Niesporek1, Reiner Umathum1, Thomas M. Fiedler1, and Armin M. Nagel1,2

1Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, 2Diagnostic and Interventional Radiology, University Medical Center Ulm, Ulm, Germany

Synopsis

Functional information of cell and tissue viability can be obtained via dynamic 17O-MRI. The quantification of H217O-concentration as a turnover product of oxidative phosphorylation in combination with 17O2-inhalation enables mapping of the cerebral metabolic rate of oxygen consumption (CMRO2). Due to low MR-sensitivity of the 17O-nucleus, spatial as well as temporal resolution is limited. Induced partial volume effects hinder accurate and stable signal quantification which is essential for dynamic studies. High resolution 17O-data with (4.5mm)3 was acquired with a Golden Angle acquisition scheme which allowed reconstruction of arbitrary time-frames. A partial volume correction algorithm was applied to test correction capability of data of different temporal resolution.

PURPOSE

Tissue and cell viability information can be accessed via evaluation of the energy balance. An interesting pathway is the oxygen metabolism which can be analyzed via oxygen-17-MRI (17O-MRI). With additional 17O2-inhalation, the measured variation of H217O-concentration in time (nat. abundance 0.037%), which is the turnover product of oxidative phosphorylation, allows localized mapping of the cerebral metabolic rate of oxygen consumption (CMRO2)1. However, the in-vivo signal of the 17O-nucleus is reduced by 105 compared to protons (1H). Additionally extremely fast transverse relaxation is caused by the electrical quadrupole interaction (I=5/2). Thus pulse-sequences that enable ultra-short echo-times and high SNR-efficiency such as 3D density adapted radial (3D-DAPR)2 or twisted projection imaging3 are required. Spherical acquisition schemes and T2*-relaxation additionally enlarge the full-widths-at-half-maximums (FWHM) of the point-spread-functions (PSF) and thus induce strong partial-volume (PV) effects which diminish the accuracy of signal evaluation of time-resolved 17O-MRI. Recently, a partial volume correction (PVC) algorithm4 was successfully applied5,6 to non-proton MRI. Spatial resolution, as well as temporal resolution for a CMRO2-experiment is essential for accurate determination of functional information. Here, PVC 17O-data with high spatial and temporal resolution, with additional B1-correction is presented as pre-evaluation for future CMRO2-experiments.

METHODS

A custom-built quadrature 17O/1H-head-coil was optimized for in-vivo measurements. Imaging was conducted on a 7T MR-system (Magnetom 7T, Siemens AG). 17O-data of a healthy volunteer (age 26) with nominal resolution of (4.5mm)3 was acquired with a 3D-DAPR-sequence with golden angle (GA)7 projection acquisition scheme (Fig.1B) which enables reconstruction of arbitrary time-frames. Additional data were acquired to calculate B1-maps (Fig.1A) with a phase-sensitive method8 (TR/TE=72ms/0.56ms, 8000 projections, Θ=75°, resolution: (4.5mm)3). 17O-data was reconstructed with a SNR-enhancing Hamming filter (FOV: 243x243x243mm3) and B1-corrected. The 1H-channel was used for shimming and acquisition of basic anatomical reference-data (3D-GRE-sequence (TR/TE=8.1ms/4.88ms, Θ=10°, (1mm)3)) that was used as registration basis. Anatomical masks for CSF, grey (GM) and white matter (WM) were obtained from high resolution anatomical data (Fig.1C) that was acquired with a 24-channel 1H-head-coil . Oxygen-data of different temporal resolution was then PV corrected to quantify the water-content considering GM, WM and two CSF compartments (lateral ventricles (CSFi)/sulci (CSFo)). Previously determined T2*-values for GM (T2*=2.5ms), WM (T2*=2.8ms), CSF (T2*=3.9ms) and H2O (T2*=5.3ms) where used for PSF-simulation. Tubes filled with pure H2O where used as external reference.

RESULTS

Quantification accuracy of 17O-signal was improved for all considered temporal resolutions where discrepancy between CSFi and CSFo (ΔCSF) was reduced. Before PVC a mean mismatch of ΔCSF=18% was cut back to ΔCSF=3% after correction. Both considered brain matter compartments (GM and BM) experienced upward signal correction, where grey matter was shifted from 52±1% to 82±1% and white matter from 50±1% to 61±2%. For low SNR-data (Δt=1min/2min, SNR=7/10) similar results with little deviation to high-SNR data (Δt=30min, SNR=36) was obtained. Signal and correction stability was evaluated considering several time steps, where 17O-signal was corrected individually (Fig.2). Corrected water-content remained stable within ±2% for brain matter compartments and ±5% for CSF with maximal fluctuation of 8%. The measured B1-map allowed localized signal correction and was capable of adjusting the signal ratio Rtubes of left and right reference (Fig.1B, 1&2) tube from Rtubes=0.77±0.03 to Rtubes=1.02±0.05.

DISCUSSION

Signal quantification of in-vivo 17O-data is strongly influenced by PV effects. With chosen GA projection acquisition scheme it was possible to evaluate data-sets with a nominal resolution of (4.5mm)3 with variable temporal resolution. The applied PVC algorithm is capable of recovering water-content values close to literature9 with even low SNR-data (Δt=1min), showing good signal and correction stability. If no correction was applied GM and WM values were underestimated by up to 35%; after correction signal mismatch was reduced to <10%. However, GM and WM values are still underestimated by 4-10%, most likely due to not fully corrected transverse relaxation. It was seen that B1-mapping with a phase sensitive method is feasible for 17O-MRI, showing good signal adaption for reference tubes where B1 heavily influences homogeneity. Additional information obtained from the built in 1H-channel improved registration accuracy of high resolution anatomical data and allowed shimming. General signal stability of non-moving calibration tubes and brain matter compartments was very good, with maximal fluctuation of 3-4%. Variation for CSF compartments was stronger due to higher sensitivity to motion which was not considered in correction. Separate anatomical mask for each time step would improve correction stability.

CONCLUSION

In this study 17O-data was evaluated considering spatial and temporal aspect of signal quantification which is of interest for dynamic 17O-studies. Applied PVC allowed correct, stable signal quantification of considered brain compartments for high spatial and temporal resolution with additional capable B1-correction.

Acknowledgements

References

1. Atkinson et al., NeuroImage 2010 (2): 723-733, 2. Nagel et al., Magn Reson Med 2009 (62):1565-73, 3. Boada et al., Magn Reson Med (1997); 37: p. 706-715, 4. Rousset et al., J Nucl Med 1998(5):904-911, 1998, 5. Niesporek et al., NeuroImage, 2015(112): 353–363 6. Hoffmann et al., MAGMA 2014(27):579-87, 7. Chan, R.W. et al., Magn Reson Med 2009(61): p. 354–363, 8. Morell, Magn Reson Med 2008 (60):889-94, 9. Neeb et al., NeuroImage, 2006 (31): 1156–1168

Figures

Fig. 1 Used data for signal evaluation: Flip-angle-map of nominal Θ=53° (A) calculated with phase-sensitive method from additional data, in-vivo 17O-image (B) (3D-DAPR-sequence with: TR/TE=12ms/0.56ms, 25000 projections, Golden Angle acquisition scheme, 6 averages, Θ=53°, nominal resolution: (4.5mm)3; TAcq=30min) with external reference tubes (1,2) and T1-weighted anatomical reference (C) (MPRAGE-sequence, (0.59mm)3).

Tab. 1 Determined water-content of lateral ventricles (CSFi), sulci (CSFo), grey (GM), white matter (WM) and H2O from (4.5mm)3-17O-data with different temporal resolution and corresponding literature values9. SNR was determined in external reference tubes and was in the range of SNR=7-36.

Fig. 2 Evaluation of stability: (A) PVC corrected signal of CSF (blue), grey (green) and white matter (orange) with Δt=1min; (B) course of signal for calibration tubes remains very stable (within 3% of mean). Signal for brain compartments show fluctuation which might be caused by non-corrected motion during measurement.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3965